Bayesian scoring of SA does not seem correct (SuSE 9.3)